CN106952241A - A kind of electromagnetic image method of partition based on morphological method and Meanshift algorithms - Google Patents
A kind of electromagnetic image method of partition based on morphological method and Meanshift algorithms Download PDFInfo
- Publication number
- CN106952241A CN106952241A CN201710287070.3A CN201710287070A CN106952241A CN 106952241 A CN106952241 A CN 106952241A CN 201710287070 A CN201710287070 A CN 201710287070A CN 106952241 A CN106952241 A CN 106952241A
- Authority
- CN
- China
- Prior art keywords
- image
- electromagnetic
- carried out
- morphological
- algorithms
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000000877 morphologic effect Effects 0.000 title claims abstract description 14
- 238000005192 partition Methods 0.000 title claims abstract description 10
- 230000005855 radiation Effects 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 7
- 238000005286 illumination Methods 0.000 claims description 7
- 230000010339 dilation Effects 0.000 claims description 3
- 230000003628 erosive effect Effects 0.000 claims description 3
- 239000013589 supplement Substances 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 abstract description 11
- 238000001914 filtration Methods 0.000 abstract description 9
- 230000001629 suppression Effects 0.000 abstract description 4
- 238000012795 verification Methods 0.000 abstract description 2
- 238000003384 imaging method Methods 0.000 description 7
- 239000013598 vector Substances 0.000 description 7
- 239000000284 extract Substances 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000011084 recovery Methods 0.000 description 4
- 238000000638 solvent extraction Methods 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000005670 electromagnetic radiation Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000005684 electric field Effects 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 230000033228 biological regulation Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 239000000047 product Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
- 238000013316 zoning Methods 0.000 description 1
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
Abstract
The present invention is the electromagnetic image method of partition based on morphological method and Meanshift algorithms, the present invention carries out homomorphic filtering noise suppression to image first, then image is handled using morphological method, so that image is easier to carry out Mean Shift cluster segmentations, the threshold parameter of image roughness and pixel value mean deviation as clustering algorithm is then calculated.Carry out MS segmentations are connect, and same signal area are determined whether to image after segmentation, merged for same signal area.Emulation and experimental verification are carried out to method, result presentation method can be carried out image noise suppression with efficiently and accurately and be split to multizone multi-frequency electromagnetic image.
Description
Technical field
The present invention relates to the noise suppressed and sectional image recovery algorithms being imaged to multizone multi-frequency electromagnetic radiation source, tool
Body is related to electromagnetic surveying and image processing field.
Background technology
At present there are many shortcomings in the detection means to electromagnetic interference source, and such as detection speed is slow, be difficult to complete detection, and
The method for carrying out magnetography using parabolic reflector can rapidly and accurately detect the position of interference source.Magnetography studies mesh
Before be concentrated mainly on passive millimeter wave imaging field, influenceed by factors such as antenna size, diffraction limiteds, institute is into image resolution ratio
Not high, the related high-resolution super resolution algorithm that carries becomes the focus of research.And the detection for electromagnetic interference source is imaged,
Wavelength is longer compared with mm-wave imaging, and diffraction limited is even more serious, and extended into the point comprising interference sources in image
How function, improve its resolution ratio and interference source clearly differentiated into new difficult point.
The content of the invention
The technology of the present invention solves problem:Overcome the deficiencies in the prior art, it is proposed that one kind based on morphological method and
The electromagnetic image method of partition of Meanshift algorithms, can make image procossing meet non-linear spy of the human eye for luminosity response
Property, it is to avoid the distortion of Fourier transform processing is directly carried out to image.
The technology of the present invention solution:Before partitioning algorithm is carried out, first image is carried out at denoising using homomorphic filtering
Reason, homomorphic filtering is a kind of image processing method that frequency filter and greyscale transformation are combined, it by image illumination/
The basis that Reflectivity Model is handled as frequency domain, the quality of image is improved using brightness range and enhancing contrast is compressed.
Next region is carried out to the image after denoising to detach.The basis that region is detached is the segmentation for carrying out image.
Simple image partition method includes even partition and non-uniformly distributed load is equal.It is horizontal into image i.e. to institute using the method for even partition
To segmentation or longitudinally split, the subregional number of regulation institute, by the segmentation of image uniform, this method be clearly it is inapplicable,
Each position of interference source can not be accurately found, and an interference source is easily assigned into different zones.Non-uniformly distributed load rule is according to each
Position of interference source and size are split, and can accurately find resolution ratio identical interference source region, but need to enter manually
OK, time-consuming and without unified dividing method.It is existing in recent years many effective as the hot issue of image processing field
Image partition method, and wherein be applied to divide electromagnetic radiation source imaging based on the method that Mean Shift are average drifting
Cut.
But Mean Shift picture portion algorithms tend to identification convexity distribution, size is close, the cluster of similar density, because
This is needed to handle original electromagnetic surveying imaging so that edges of regions to be split is tried one's best smoothly, and is convexity distribution, so that
Improve the accuracy of Mean Shift algorithms.The image obtained after over-segmentation, which has, isolates the same area or overlapping possibility,
Therefore the judgement that image is isolated for overlay chart picture and mistake is increased after the calculating of Mean Shift algorithms, final point obtained
Cutting image accurate can go out regional respectively.
The present invention comprises the following steps:
Step 1, obtain the degraded image of unknown radiation source, logarithm operation carried out to image, convert the image into illumination and
The form that two parts of reflection are added, image after being taken the logarithm;
Step 2, to taking the logarithm after image be filtered processing, filter the high-frequency noise of image, so as to force down brightness range,
Again to image fetching number, recover electromagnetic power image, obtain image after denoising;
Step 3, sorbel edge extractings are carried out to image after denoising, and edge carried using morphological image method
Image is taken to carry out dilation erosion operation, so as to be the overall bianry image that signal area is convexity edge by image completion;
Step 4, the roughness of overall bianry image is calculated, roughness represents the size of average texture in image, as
Distance clusters threshold value hs in MeanShift algorithms, then calculates image pixel average offset value, this each picture of deviant representative image
Pixel value situation of change between vegetarian refreshments, threshold value h is clustered as pixel value in MeanShift algorithmsr;
Step 5, according in step 4 calculate obtain apart from clustering parameter hsWith pixel value clustering parameter hrTo blank map picture
Carry out MeanShift algorithms and carry out image block;
Step 6, according to the piecemeal situation in step 5, the different regions of image after step 2 denoising are extracted out signaling zone
Domain, then block image supplement is original image size, convenient to recover followed by image;
Step 7, the image that piecemeal is obtained is judged, if same signal segmented areas is then merged.
Advantages of the present invention is with good effect:
(1) for needing, the marginal definition in each piece of region of piecemeal is stricter to be wanted traditional images block algorithm
Ask, it is poor for the limb recognition and image segmentation of the fuzzyyer image in edge.And obtained for magnetography
Image for, due to point spread function influence its electric-field intensity distribution trend close to sinc functions, i.e., in edge
Pixel value changes are slow, thus the resolving effect of electric-field intensity image that is obtained for magnetography of traditional images block algorithm compared with
Difference, the accuracy of subregion identification is very low.And after first being handled using morphological method, the signal area of magnetography can be transformed to
Edge is obvious and essentially convexity edge image region, and MeanShift clustering algorithms are calculated after being more easy to.
(2) because electric magnetic image is imaged for wideband electromagnetic, with the reduction of electromagnetic signal frequency, electric magnetic image
Point spread function main lobe width increases its imaging region size in image plane and also gradually increased, therefore comes for block algorithm
Say, the clustering parameter of its piecemeal needs to be adjusted according to signal area size.In the present invention, calculate image roughness and
Pixel value mean deviation is used as the threshold parameter of clustering algorithm, and the two parameters are big in the pixel value region for having weighed image
Small, change speed etc. can embody the texture situation of image, therefore use it as the parameter of clustering algorithm to a certain extent
The purpose of broad band region segmentation can be reached to a certain extent.
Brief description of the drawings
Fig. 1 realizes flow chart for the inventive method;
Fig. 2 is different frequency multi radiation sources electromagnetic image under noise-free case;
Fig. 3 is different frequency multi radiation sources electromagnetic image progress sobel edge-detected images under noise-free case;
Fig. 4 is the bianry image after morphological image method is handled
The mono signal image that Fig. 5 is obtained after Meanshift partitioning algorithms, wherein a is image after signal 1 is extracted, and b is
Signal 2 extracts image, and c is that signal 3 extracts image;
Fig. 6 electric magnetic images experiment obtain it is noisy in the case of different frequency multisignal source electromagnetic image;
Low noise images of the Fig. 7 after homomorphic filtering;
The mono signal image that Fig. 8 is obtained after Meanshift partitioning algorithms, wherein a is image after signal 1 is extracted, and b is
Signal 2 extracts image;
Fig. 9 mono signals image passes through LR superresolution processings, merges image after multiple signal super-resolution and obtains final high-resolution
Rate electromagnetic image.
Embodiment
This method is described in further detail below in conjunction with drawings and Examples.
As shown in figure 1, the present invention is directed to the problem of same magnetography image has different frequency radiation source, one is proposed
Plant the wideband electromagnetic image segmentation algorithm being combined using morphology and MeanShift clustering algorithms.Image enters after to taking the logarithm
Row filtering process, filters the high-frequency noise of image, so as to force down brightness range, then to image fetching number, recovers electromagnetic power figure
Picture.
Step 1, obtain the degraded image of unknown radiation source, logarithm operation carried out to image, convert the image into illumination and
The form that two parts of reflection are added.
That is image f (x, y) can be expressed as illumination i (x, y) and reflection r (x, y) two-part product:
F (x, y)=i (x, y) r (x, y)
Therefore illumination and reflecting part can not be operated respectively respectively for original image, therefore selection is entered to image
Row logarithm operation, it is assumed that z (x, y)=lnf (x, y)=lni (x, y)+lnr (x, y), to taking the logarithm, image carries out Fourier's change
Change, then
I.e.:
Z (u, v)=Fi(u,v)+Fr(u,v)
Wherein u, v are two-dimentional null tone rate, and Z is the Fourier transformation of original image, and Fi is the Fourier transformation of illumination image,
Fr is the Fourier transformation of reflected image.
Image is handled if by a filter function H, then can preferably remove the multiplicative noise in image.
Step 2, sorbel edge extractings are carried out to image after denoising, and edge carried using morphological image method
Image is taken to carry out dilation erosion operation, so as to be the bianry image that signal area is convexity edge by image completion.
Step 3, the roughness of general image is calculated, roughness represents the size of average texture in image, can conduct
Distance cluster threshold value hs ideal chose in MeanShift algorithms.Image pixel mean deviation is calculated, this value representative image is each
Pixel value situation of change between pixel, can cluster threshold value hr ideal chose as pixel value in MeanShift algorithms.
The specific method that roughness is calculated is as follows.First to each pixel, the average value for defining its neighbours' window is Mk
(x, y), namely:
Wherein f (i, j) represents image in the pixel value of point (i, j), k=1,2 ... ..., L, wherein 2k×2kIt is maximum for image
Window, then L value can be obtained by following formula:2L≤ B < 2L+1, B is the length of the most short side of image.
Then the nonoverlapping maximum difference of calculation window, Ek,h(x, y), Ek,v(x, y), Ek,d(x, y), wherein Ek,h(x,y)
Represent the maximum difference of horizontal direction, Ek,v(x, y) represents the maximum difference of vertical direction, Ek,d(x, y) represents diagonally opposed
Maximum difference.The calculation of this three is as follows:
Ek,h(x, y)=| Mk(x+2k-1,y)-Mk(x-2k-1,y)|
Ek,v(x, y)=| Mk(x,y+2k-1)-Mk(x,y-2k-1)|
Ek,d(x, y)=| Mk(x+2k-1,y+2k-1)-Mk(x-2k-1,y-2k-1)|
Then the maximum of the E corresponding to each k is obtained, namely:
To each pixel, the optimal texture size W of image-region is obtainedbest(x, y)=2k.Finally, all W are calculatedbest's
The average value of sum, as roughness F:
The specific calculation of pixel mean deviation amplitude is as follows, and image size is M × N (pixel number), and P is some position
Pixel value size is put, A is the pixel average of whole image, the pixel mean deviation amount of whole imageThen:
Step 4, according in step 4 calculate obtain apart from clustering parameter hs and pixel value clustering parameter hr to blank map picture
Carry out MeanShift algorithms and carry out image block.
Mean Shift algorithms basic thought arbitrarily chooses a point, work is put with this wherein to assume a d dimension space
For the centre of sphere, make the higher-dimension ball that radius is h.Fall the vector that can all be produced a little between the center of circle in ball, vector using the center of circle as
Starting point.By all addition of vectors in ball, addition result is Mean shift vectors.With the vectorial terminals of the Mean Shift of gained
As new starting point, step before continuation, final vector can converge on probability density maximum.
For probability density function f (x), in known n sampled point xi, f (x) kernel function estimation is:
Wherein h is the radius of higher-dimension ball, and d is Spatial Dimension, w (xi) >=0 is sampled point xiWeight, K (x) is kernel function,
Kernel function need to meet ∫ k (x) dx=1.
Definition kernel function K (x) profile function k (x), i.e. K (x)=k (| | x | |2), define g (x) and led for the negative of k (x)
Number, i.e. g (x)=- k'(x), then corresponding kernel function G (x)=g (| | x | |2)。
Probability density function f (x) gradient is asked to estimate i.e.Estimation:
By g (x)=- k'(x), G (x)=g (| | x | |2), above formula can be written as:
It is to be right using kernel function G (x) in second bracket of above formula in Mean Shift vectors, first bracket
F (x) estimation, therefore can be abbreviated as:
Wherein:
This shows the maximum descent direction of the direction next -event estimator density gradient of Mean Shift vectors, i.e., have at convergence point
For a sub-picture, pixel is uniformly distributed on image, in the absence of the density of point.How pixel is defined
Probability density turns into key.
Define certain pixel x probability density:It is in d higher-dimension balls to falling in dimension using h as radius using x as the center of circle
Point xiFormulate following rule:
(1) pixel x value (i.e. the modulus value of field intensity in magnetography) and pixel xiValue closer to, define x points it is general
Rate density is higher.
(2)xiWith x position closer to definition x point probability density is higher.
It is for the Multilayer networks formula that certain in piece image is put then:
Wherein xLRepresent the size of pixel point value, xCRepresent pixel position coordinates, hLFor the pixel value higher-dimension radius of a ball, hC
For apart from the higher-dimension radius of a ball.
Can obtain to a certain sub-picture carry out Mean-Shift segmentations iterative formula be:
For parabolic reflector institute into interference source images, randomly select any point therein and carry out Mean according to above formula
Shift iteration, sets convergence threshold, and its convergence point is recorded after convergence, then randomly selects left point, until traversal institute is a little.Will
Converge on the close convergence point of the close pixel value in position is attributed to a little same cut zone.
Step 5, according to the piecemeal situation in step 5, the different region of image after step 2 denoising is extracted out signal area,
Then block image supplement is original image size, convenient to recover followed by image.
Step 6, the image that piecemeal is obtained is judged, if same signal segmented areas is then merged.
Judge the position where each segmented areas, if there is two pieces of regions overlapping range more than 70%, then then
Judge that two regions, for identical signal cut zone, are merged, under the minimum region for taking two regions to two regions
Boundary and maximum region of upper bound, the region merged.
Invention is emulated below, so that three different frequency radiation sources are present in a width electromagnetic image simultaneously as an example,
The frequency in one of source is 1GHz, and the frequency in two other source is 3GHz.On the object plane at 5 meters of electric magnetic image
These three sources, by the electric magnetic image of diffraction limited, obtained in image plane after point spread function effect is degenerated
Image.System, which is used, is distributed 201 × 201 sampled points on 1 meter × 1 meter of imaging surface, imaging surface.Emulate obtained electromagnetic image
As shown in Fig. 2 being different frequency multi radiation sources electromagnetic image under noise-free case.Because emulating image does not have noise substantially, because
The step of this omits homomorphic filtering denoising.Image such as Fig. 3 after sobel rim detections is carried out to image, Fig. 3 is under noise-free case
Different frequency multi radiation sources electromagnetic image carries out sobel edge-detected images, carries out result such as Fig. 4, Fig. 4 after Morphological scale-space and is
Bianry image after the processing of morphological image method, three regions, such as Fig. 5, Fig. 5 are obtained after carrying out Mean Shift cluster subregions
For the mono signal image obtained after Meanshift partitioning algorithms, wherein a is image after signal 1 is extracted, and b extracts for signal 2
Image, c is that signal 3 extracts image, it can be seen that method is relatively good to be carried out to multiple different frequency radiation source electromagnetic images
Multidomain treat-ment, carries out image reconstruction to each subregion after being and lays the first stone.
Followed by specific experiment checking, using two electromagnetic horns as electromagnetic interference source, frequency is respectively 1.8GHz
And 3GHz, it is imaged using electric magnetic image, obtained electromagnetic image is Fig. 6, the experiment of Fig. 6 electric magnetic images is contained
Different frequency multisignal source electromagnetic image in the case of making an uproar, it can be seen that noise uses homomorphic filtering denoising and then removal than larger
Direct current step in image, obtained denoising image such as Fig. 7, Fig. 7 is the low noise image after homomorphic filtering, then by shape
State handles and carries out the processing of Mean Shift algorithms partitions, and obtained sectional image such as Fig. 8, Fig. 8 are split by Meanshift
The mono signal image obtained after algorithm, wherein a is image after signal 1 is extracted, and b is that signal 2 extracts image, it can be seen that image point
Cut effect relatively good.Then choose corresponding point spread function and LR iterative image recoveries are carried out to partitioned area, then by after recovery
Region merging technique obtains finally recovering region such as Fig. 9, and Fig. 9 is that mono signal image passes through LR superresolution processings, merges multiple signals and surpasses
Image obtains final high-resolution electromagnetic image after resolution, it can be seen that clearly two radiate point source.
In a word, the present invention when electromagnetic survey system obtain in electromagnetic radiation image comprising multiple frequencies and multiple distances compared with
During remote signal area, if directly carrying out image recovery and image reconstruction using identical point spread function to image, obtain
Reconstruction image effect it is poor, the noise more than comparison is had in image and some signals can be by as noise suppressed, it is therefore desirable to
The signal area of each in image is separated, carrying out different images to different zones recovers.For this zoning requirements, the present invention
Homomorphic filtering noise suppression is carried out to image first, then image handled using morphological method, so that image is easier to
Mean Shift cluster segmentations are carried out, image roughness and pixel value mean deviation is then calculated and joins as the threshold value of clustering algorithm
Number.Carry out MS segmentations are connect, and same signal area are determined whether to image after segmentation, carried out for same signal area
Merge.Emulation and experimental verification are carried out to invention, result presentation method can be with efficiently and accurately to multizone multi-frequency electromagnetic
Image carries out image noise suppression and split.
Above example is provided just for the sake of the description purpose of the present invention, and is not intended to limit the scope of the present invention.This
The scope of invention is defined by the following claims.The various equivalents that do not depart from spirit and principles of the present invention and make and repair
Change, all should cover within the scope of the present invention.
Claims (1)
1. the electromagnetic image method of partition based on morphological method and Meanshift algorithms, it is characterised in that realize step such as
Under:
Step 1, the degraded image of unknown radiation source is obtained, logarithm operation is carried out to image, illumination and reflection is converted the image into
The form that two parts are added, image after being taken the logarithm;
Step 2, to taking the logarithm after image be filtered processing, the high-frequency noise of image is filtered, so as to force down brightness range, then right
Image fetching number, recovers electromagnetic power image, obtains image after denoising;
Step 3, sorbel edge extractings are carried out to image after denoising, and using morphological image method to edge extracting figure
As carrying out dilation erosion operation, so as to be the overall bianry image that signal area is convexity edge by image completion;
Step 4, the roughness of overall bianry image is calculated, roughness represents the size of average texture in image, as
Distance clusters threshold value hs in MeanShift algorithms, then calculates image pixel average offset value, this each picture of deviant representative image
Pixel value situation of change between vegetarian refreshments, threshold value h is clustered as pixel value in MeanShift algorithmsr;
Step 5, according in step 4 calculate obtain apart from clustering parameter hsWith pixel value clustering parameter hrBlank map picture is carried out
MeanShift algorithms carry out image block;
Step 6, according to the piecemeal situation in step 5, the different regions of image after step 2 denoising are extracted out signal area, so
Block image supplement is original image size afterwards, convenient to recover followed by image;
Step 7, the image that piecemeal is obtained is judged, if same signal segmented areas is then merged.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710287070.3A CN106952241B (en) | 2017-04-27 | 2017-04-27 | Electromagnetic image blocking method based on morphological method and Meanshift algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710287070.3A CN106952241B (en) | 2017-04-27 | 2017-04-27 | Electromagnetic image blocking method based on morphological method and Meanshift algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106952241A true CN106952241A (en) | 2017-07-14 |
CN106952241B CN106952241B (en) | 2020-03-10 |
Family
ID=59476817
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710287070.3A Active CN106952241B (en) | 2017-04-27 | 2017-04-27 | Electromagnetic image blocking method based on morphological method and Meanshift algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106952241B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112418147A (en) * | 2020-12-02 | 2021-02-26 | 中国人民解放军军事科学院国防科技创新研究院 | Track \30758identificationmethod and device based on aerial images |
CN114710215A (en) * | 2022-04-08 | 2022-07-05 | 郑州大学 | Method for fast blind detection of frequency hopping signal |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101599172A (en) * | 2008-06-06 | 2009-12-09 | 上海吉量软件科技有限公司 | The illumination compensation splitting method of the text image of inhomogeneous illumination |
CN104949981A (en) * | 2014-03-24 | 2015-09-30 | 华中科技大学 | Automatic detection method and system for cotton five-euphylla period |
US20170068855A1 (en) * | 2015-07-17 | 2017-03-09 | Tata Consultancy Services Limited | Method and system for facilitating real time detection of linear infrastructural objects by aerial imagery |
CN107133965A (en) * | 2017-04-17 | 2017-09-05 | 商丘工学院 | One kind is based on computer graphic image morphological image segmentation method |
-
2017
- 2017-04-27 CN CN201710287070.3A patent/CN106952241B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101599172A (en) * | 2008-06-06 | 2009-12-09 | 上海吉量软件科技有限公司 | The illumination compensation splitting method of the text image of inhomogeneous illumination |
CN104949981A (en) * | 2014-03-24 | 2015-09-30 | 华中科技大学 | Automatic detection method and system for cotton five-euphylla period |
US20170068855A1 (en) * | 2015-07-17 | 2017-03-09 | Tata Consultancy Services Limited | Method and system for facilitating real time detection of linear infrastructural objects by aerial imagery |
CN107133965A (en) * | 2017-04-17 | 2017-09-05 | 商丘工学院 | One kind is based on computer graphic image morphological image segmentation method |
Non-Patent Citations (3)
Title |
---|
徐国保: ""一种用细胞神经网提取遥感图像边缘的新方法"", 《计算机应用研究》 * |
杜鑫: ""一种电磁干扰源成像多分辨率分区算法"", 《强激光与粒子束》 * |
熊平,等: ""带宽自适应Mean Shift 图像分割算法"", 《计算机工程与应用》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112418147A (en) * | 2020-12-02 | 2021-02-26 | 中国人民解放军军事科学院国防科技创新研究院 | Track \30758identificationmethod and device based on aerial images |
CN114710215A (en) * | 2022-04-08 | 2022-07-05 | 郑州大学 | Method for fast blind detection of frequency hopping signal |
CN114710215B (en) * | 2022-04-08 | 2024-02-02 | 郑州大学 | Method for fast blind detection of frequency hopping signal |
Also Published As
Publication number | Publication date |
---|---|
CN106952241B (en) | 2020-03-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108154519B (en) | Method and device for segmenting blood vessels in fundus image and storage medium | |
Fathi et al. | Automatic wavelet-based retinal blood vessels segmentation and vessel diameter estimation | |
CN107085708B (en) | High-resolution remote sensing image change detection method based on multi-scale segmentation and fusion | |
CN103279957A (en) | Method for extracting remote sensing image interesting area based on multi-scale feature fusion | |
CN103745237A (en) | Face identification algorithm under different illumination conditions | |
CN104537622B (en) | The method and system that raindrop influence is removed in single image | |
CN108062508B (en) | The extracting method of equipment in substation's complex background infrared image | |
Mustafa et al. | Image enhancement technique on contrast variation: a comprehensive review | |
CN101833658A (en) | Illumination invariant extracting method for complex illumination face recognition | |
Asmuni et al. | An improved multiscale retinex algorithm for motion-blurred iris images to minimize the intra-individual variations | |
CN104036461B (en) | A kind of Infrared Complex Background suppressing method based on Federated filter | |
CN106952241A (en) | A kind of electromagnetic image method of partition based on morphological method and Meanshift algorithms | |
CN108932492A (en) | A kind of image fingerprint extracting method based on non-sampled shearing wave conversion | |
Chandra et al. | A survey on advanced segmentation techniques in image processing applications | |
Wu et al. | Research on crack detection algorithm of asphalt pavement | |
CN108596928A (en) | Based on the noise image edge detection method for improving Gauss-Laplace operator | |
TWI460667B (en) | Rebuilding method for blur fingerprint images | |
CN115797773A (en) | Ultra-wideband radar target identification method based on edge detection and region segmentation | |
CN115147613A (en) | Infrared small target detection method based on multidirectional fusion | |
Shen et al. | Orthogonal moments and their application to motion detection in image sequences | |
Chen et al. | An edge detection method for hyperspectral image classification based on mean shift | |
Wang et al. | Edge detection from high-resolution remotely sensed imagery based on Gabor filter in frequency domain | |
Rodrigues et al. | Towards breast ultrasound image segmentation using multi-resolution pixel descriptors | |
Duman et al. | Target detection in sar images using codifference and directional filters | |
Zhang et al. | Infrared small target detection using modified order morphology and weighted local entropy |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |